The Chess Match That Changed Our Minds About AI

In May of 1997, Garry Kasparov sat down at a chess board in a Manhattan skyscraper. Kasparov, considered the best chess player of all time, wasn’t challenging another grandmaster. He was playing with an AI called Deep Blue. Deep Blue was one of the world’s most powerful supercomputers, built by IBM with a specific goal in mind: to beat humanity at its own game. For IBM, billions of dollars worth of business clout was on the table, and to a certain extent, Kasparov was playing for the fate of chess itself. He had never lost a multi-game match in his entire career. Could a machine beat him? Newsweek ran a cover story with his picture alongside the words “The Brain’s Last Stand.” As Kasparov joked years later, “No pressure.”

Over 25 years later, we’re living through another moral panic over artificial intelligence. Thanks to ChatGPT, once hypothetical questions about the future of work, art, and disinformation are now immediate concerns. The only question is how far AI can go. Google CEO Sundar Pichai offered his expectations a few years ago. AI is “one of the most important things that humanity is working on,” he said. “It’s more profound than, I don’t know, electricity or fire.” It seems we’re dealing with something entirely new—except we aren’t.

In the ‘90s, we went down a similar road, with the same questions, the same fears, and nearly identical conversations. The biggest face-off between man and machine already happened, and it culminated with a single move during a few games of chess. As the world watched Kasparov stare out at a field of black and white pieces, we got our first glimpse of what it feels like when computers start acting like human beings.

“There are very few instances of an arena where the human body and mind can compete on equal terms with a computer or a robot,” Kasparov said in a 2017 TED Talk. “It was my blessing, and my curse, to literally become the proverbial man in the man versus machine competition.”

It’s a story of cheating accusations, media frenzy, and a final answer to one of life’s great questions: have we created machines so powerful that they can replace us? And if so, what now?

“If you want to know what the future of AI looks like, look at chess,” Frederic Friedel, the co-founder of ChessBase and Kasparov’s computer advisor during the match, told Gizmodo. “It happened to us first, and it’s going to happen to all of you.”

The Contenders

Garry Kasparov wasn’t just a chess savant, he was a cultural phenomenon. He was famous enough to be in a Pepsi commercial. At the height of his stardom, the name “Kasparov” was sort of like “Einstein,” marking the pinnacle of human intelligence. He was rebellious and outspoken, and Kasparov knew how to work the media to his advantage. He’d stunned the chess intelligentsia a decade prior, defeating grandmaster Anatoly Karpov at just 22. That made him the youngest-ever world champion, a record that’s still unbroken.

In 1997, Kasparov was a darkly handsome 34-year-old with piercing eyes and a perpetual look of seriousness. Though he stood at 5’9’’, Kasparov was so intimidating on the chessboard that some opponents said it felt like he towered over them.

“He was a monster,” Friedel said. Kasparov was unstoppable, leagues ahead of all the other grandmasters. “He’s one of the deepest players I’ve ever encountered.”

IBM, on the other hand, wasn’t at the top of its game. Through the 1980s, IBM was one of the most powerful companies on earth, manufacturing 80% of the computers in the United States. But IBM’s dominance slipped away to Microsoft and other competitors as the 20th century wore on. The company wanted to prove it was still a leader, and building the world’s smartest computer was a perfect PR coup.

“For IBM, I think it was an indication that we could sort of swing for the fences,” said Murray Campbell, an AI researcher at IBM and one of the lead architects behind Deep Blue. “It changed the mindset in terms of how the world reacted to IBM. For many people, it was their first experience of a computer doing something they thought only humans were capable of.”

The games were actually a rematch. Kasparov beat Deep Blue, just barely, in a series of games in 1996. But the computer won the first game, and two out of six were a draw. IBM wanted more, and Kasparov was excited about the scientific pursuit, so they agreed to play again. It wasn’t just about bragging rights, either. The loser of the match would take home $400,000, while the winner would net $700,000. The difference meant $300,000 was at stake during the games, not a small amount of cash. But there was even more money at stake. In the weeks after the 1996 match, IBM’s stock rose almost 20%.

‘’Forget the $300,000,’’ said Maurice Ashley, an international grandmaster, speaking to the New York Times. “The future of humanity is on the line.’’

Weighing in at 1.4 tons, Deep Blue was a pair of two hulking computer towers, each over six-and-a-half feet tall. IBM upgraded the machine for the 1997 games, with over 500 processors and 480 specially-designed “chess chips’’ running in parallel. Deep Blue could search through hundreds of millions of chess positions a second to find the best move.

“As soon as computers know how to do something, people say it’s not intelligence anymore, it’s just an algorithm. But I think that’s not really fair,” Campbell said. “Deep Blue was intelligent—if only a tiny bit. But that’s just the bit it needed to handle this problem.”

At the time, no system outside of the human mind could approach Deep Blue’s chess abilities. If Deep Blue could beat Kasparov, it would be the best chess player on earth.

The Anti-Computer Strategy

Kasparov arrived in New York in May with his posse just before the match. They stayed at the Plaza Hotel, eating meals as a team and taking walks together through Central Park. “Those were not happy days,” Kasparov said in an interview years later. “From day one we had so much tension.”

Kasparov didn’t have access to any of Deep Blue’s prior games, which put the grandmaster at a significant disadvantage. “It was very intimidating for him,” Friedel said. “Any top player will prepare for their opponent, but this was a blind game for Garry.”

Still, Kasparov was the favorite. “The feeling in the room was Kasparov was the guaranteed winner,” said Monty Newborn, former chairman of the Computer Chess Committee and head of officials for the match.

The games took place on the 35th floor of the Equitable Building in Manhattan. Only a few dozen VIPs were allowed in the room, but on the floors below, hundreds of spectators packed an auditorium to watch the games play out on screen, with grandmasters commentating on stage.

During the games, rotating members of the IBM team typed in Kasparov’s moves and moved the pieces on Deep Blue’s behalf. Murray Campbell was one of them.

“It was the thrill of my life,” Campbell said. “I was nervous going into the first match in ‘96 because the system was brand new and we hadn’t really gotten the chance to thoroughly test it. But by 1997, my expectations were greatly increased.”

The grandmaster drew white, which meant he went first in game one. Kasparov moved his knight out on the board, and Deep Blue responded with its queen’s pawn, which developed into a set of moves known as the Réti opening. For such an unusual match, things we’re off to a normal start.

Deep Blue had analyzed a large history of recorded games, and it was intimately familiar with the norms of chess playing. That meant the machine was prepared for Kasparov’s usual aggressive style. But Kasparov landed on an “anti-computer” strategy.

“Anti-computer chess means you play rope-a-dope chess,” said Kenneth Regan, a computer science professor at the University of Buffalo renowned for his chess research. A computer’s calculating ability gave it an advantage when it came to attacking or defending over the course of one or two moves; it could always find the perfect response. But long-term strategy requires a different kind of “thinking.” Strategy is where great human players excel, but no computer had ever shown an ability to plan ahead for later in the game.

“So you play conservatively,” Regan said. “You don’t go into the attack right away because the computer is going to be able to calculate your attack.” In other words, you try to do as little as possible for as long as possible—but if you’re too timid, that can backfire.

That’s exactly what Kasparov did. He held back, dragging the computer across the board until he was poised to strike. It was a risky strategy.

“He was playing unorthodox chess, sometimes breaking with all known theory,“ Friedel said. Kasparov worked to make moves that threw off Deep Blue, and it kept the computer on its digital toes, but this style of play was just as foreign to Kasparov.

But as the game wore on, his anti-computer playing was lethal. Kasparov wiped Deep Blue off the board. Kasparov pinned Deep Blue’s king between his knight and his rook, and on the 45th move, the computer resigned.

The robot lost. In fact, Deep Blue played so poorly that it seemed it was going haywire. Things were off to a great start for humanity’s champion. But in the second game, everything changed.

The Chess Move That Changed History

The dance between chess and computers goes back to the very beginning. Alan Turing wrote the first chess program in 1948—before there was even a computer powerful enough to run it.

“Chess requires all sorts of forms of intelligence, reasoning, careful planning through sequences, evaluation of consequences,” Campbell said. “If we could get computers to do all of that, then we presumed computers to be intelligent.” It would be a sign that computers had crossed over.

Chess is an ideal programming challenge. There’s a record of games to analyze and the rules are clearly defined. Better still, you can break every move down into statistics based on the likelihood it leads to a win. But unlike in poker and some other complex games, the best human players aren’t doing math. It comes down to logic, strategy, knowledge of past games, and feeling out your opponent’s style. To be good at chess, you have to possess some form of intelligence—even if you’re a machine.

After losing the first game, the IBM team stayed up late into the night fixing bugs and adjusting Deep Blue’s code.

“I didn’t pay any attention because I said ‘come on, I would beat the machine whatever happens because look at game one?’” Kasparov said years later. “It’s a machine. At the end of the day, it’s stupid.”

But whatever IBM did the night before, it worked. He continued his anti-computer play, staying locked in position to try and force the computer to shuffle pieces around aimlessly. But Deep Blue wasn’t playing like a computer. It was performing like a grandmaster.

On move 36, Kasparov laid a trap. It was a calculated sacrifice. He gave Deep Blue an opening to capture two pawns, but it would give Kasparov a strong counterattack later in the game. He had played against computers before, and he knew they always focused on short-term advantages. To see his scheme coming, Deep Blue would have to do something that was impossible for a computer. It would have to do something that looked a lot like thinking.

A full 15 minutes went by while Deep Blue processed. The tension built, but Kasparov knew what was coming. It was going to move a queen and attack the pawns. But that’s not what happened. Deep Blue ignored the sacrifice and moved out a pawn instead. Not only did the machine thwart Kasparov’s attack, but it also made a move that crippled his game, setting the computer up for a complex, multi-step victory. It was a masterstroke.

Downstairs, the experts commenting on the game in the auditorium were thrilled, exclaiming that “any human grandmaster would have been proud” to play Deep Blue’s move. But Kasparov was dumbfounded.

He had just seen one of the most advanced technological feats of all time. If it takes intelligence to play great chess, then move 36 proved that Deep Blue had it. Kasparov wouldn’t accept this reality for years to come. Nine moves later, it looked hopeless. Kasparov resigned. The score was 1-1.

Paranoia Sets In

Kasparov was furious. Move 36 was too sophisticated for a computer, and how was it possible that the machine could play so well after floundering in game one? “There’s one explanation, there was a human operator behind the machine,” Kasparov said later. IBM must have used a human player to step in and tell Deep Blue what to do. They must have cheated.

After each game, Kasparov and the IBM team would walk into the auditorium, where hundreds of journalists and spectators would cheer for the grandmaster. After game two, a dispirited Kasparov skipped the conference, but when he returned to meet the crowd days later, his words were venomous.

On stage, Kasparov openly implied IBM was cheating, referencing a famous soccer match where a player used his hands to score the winning point. “Mr. Kasparov, are you saying there might have been some sort of human intervention in this game,” a reporter asked. “Well, it reminds me of Maradona’s goal against England 1986,” Kasparov replied. “He said it was the ‘hand of God.’”

The IBM team stood by as Kasparov slung accusations, waiting for their turn at the microphone. “I think what was happening here is something Garry couldn’t explain was in the machine,” said Joel Benjamin, an American chess grandmaster who worked with IBM. “And because he couldn’t explain it, he said there must be something funny going on.”

But game two’s humiliations were bigger than the loss. After the match, Friedel realized that Kasparov made a mistake. He resigned too early and missed an opportunity to drag the game out to a draw. Kasparov was so blinded by the computer wizardry that he’d thrown the game away.

Over the course of the next week, the Kasparov camp was increasingly convinced that something untoward was going on. Friedel noticed a glint from a window in a building across the street from Kasparov’s hotel room and realized it was someone pointing a camera at them. Was it paparazzi, or IBM spies? The team noticed that one of the bodyguards IBM arranged for Kasparov spoke Russian. Did IBM plant them to listen in on the Azerbaijani grandmaster?

“I think we were all a little paranoid,” Friedel said. The stress was affecting Kasparov’s game, and he was already out of his comfort zone with the anti-computer playing.

Kasparov became obsessed. He said he couldn’t focus throughout the rest of the match as his mind continuously drifted back to that impossible move in game two. Observers said he was playing far below his level. Kasparov fought Deep Blue to a draw in the next three games, looking more beleaguered each day.

Not everyone was convinced about the cheating. “Who could IBM possibly find that would play better than Deep Blue itself? Deep Blue’s playing was no surprise to me” Newborn said.

“Obviously the conspiracy theories regarding cheating and spying are ridiculous,” Campbell said.

In 1997, it was unimaginable to a lot of people that a machine could perform like Deep Blue. But IBM had spent millions of dollars and the better part of a decade developing the machine. Some of the brightest minds in computer science and artificial intelligence were working on the project. IBM even brought in a grandmaster to fine-tune the AI, who helped program responses for the exact kind of anti-computer play Kasparov was using. 25 years later, it’s unsurprising that a computer can play incredible chess, but Kasparov wasn’t ready to see that, and neither was much of the chess world.

The grandmaster continued with sly accusations in post-game press conferences, whipping the crowd into a frenzy. At one point, the audience booed IBM. Kasparov couldn’t let go of the idea that the company was cheating. A few years after the match, Kasparov starred in a documentary called Game Over: Kasparov and the Machine. It’s a scathing hour-and-a-half argument that IBM cheated.

The score was tied going into the sixth match. Kasparov was tired. Early in the game, he made one last-ditch anti-computer attempt. He launched his knight way out into Deep Blue’s territory, a risky move with a well-established response. He was gambling the computer wouldn’t know how to react. He bet wrong. In footage of the game, you see as he realizes his mistake following Deep Blue’s next move. Kasparov sunk his head into his hands. It was all over.

Later, he would tell the New York Times he felt the game was over before it started. “I was not in the mood of playing at all,” Kasparov said. “I’m a human being. When I see something that is well beyond my understanding, I’m afraid.”

After just 19 moves, Kasparov resigned, storming out of the room without the customary handshake. It was the first official defeat in Kasparov’s career. Deep Blue won. Humanity lost.

Living With the Machines

While the game was official, losing to a computer didn’t cost Kasparov his world title. Kasparov challenged IBM to a third match to settle the score for the record books. IBM declined. Kasparov never got the opportunity to save face.

“We had accomplished what we set out to do,” Campbell said.

IBM dismantled Deep Blue after the games. It was split up and donated to museums, and never played another official match. Kasparov retired from competitive chess in 2005. Today, he’s a writer and political commentator—an outspoken critic of Russian President Vladimir Putin.

In 2017, a decade after he accused IBM of cheating, Kasparov gave a TED Talk, titled Don’t fear intelligent machines. Work with them. Kasparov admitted he was wrong.

“When chess computers finally came of age, I was Mt. Everest, and Deep Blue reached the summit.” Kasparov said. “I should say of course, not that Deep Blue did it, but its human creators—Anantharaman, Campbell, Hoane, Hsu. Hats off to them. As always, machine’s triumph was a human triumph, something we tend to forget when humans are surpassed by our own creations.”

Ironically, we’ve seen a complete reversal in the chess world. In 1997, Deep Blue played so well that people thought it must be cheating with the help of a human. But in the most recent chess scandal, a player named Hans Niemann did so well in a match against world champion Magnus Carlsen that people said he must be cheating with the help of a computer.

Deep Blue didn’t end competitive chess, nor did its offspring. Modern chess engines are so good it’s assumed the best human players won’t stand a chance. You can get a chess app on your phone that makes Deep Blue look like an abacus. But people still didn’t, the same way we didn’t stop running races after the invention of the car. In fact, chess is more popular today than ever in history. There’s magic to the articulations of the human mind. Computers can’t take that away, no matter how brilliant they get.

As AI presents us with an uncertain future, this chapter of chess history can be a model for a path forward. AI is now an indispensable tool for learning to be a great chess player; there are even tournaments where human players can use AI chess engines during the match.

AI will surpass human ability in a variety of areas. It will eliminate the need for some jobs, and create some others. There’s trouble ahead. Giant corporations will take advantage, and people will be hurt along the way. But the world is evolving, not ending.

“Doomsaying has always been a popular pastime when it comes to technology,” Kasparov said. “What I learned from my own experience is that we must face our fears if we want to get the most out of our technology, and we must conquer those fears if we want to get the best out of our humanity.”